PRIVILEGED AND CONFIDENTIAL — ATTORNEY WORK PRODUCT
DEPOSITION PREPARATION OUTLINE
DR. AMARA OSEI
Named Inventor, U.S. Patent Nos. 10,891,234 and 11,045,678
C&H Matter Number: 1014-00003
Purpose of This Outline. This document is a confidential attorney work product prepared for the sole purpose of assisting in the preparation of Dr. Amara Osei for her claim-construction-phase deposition in the above-captioned matter. Dr. Osei is a Principal Engineer at Optiwave Semiconductor Corp. ("OWS" or "Optiwave") and a named inventor on U.S. Patent No. 10,891,234 ("the '234 Patent") and U.S. Patent No. 11,045,678 ("the '678 Patent"). Pursuant to the schedule established at the Initial Case Management Conference held on August 15, 2022, before Hon. Edward J. Torvelis, all claim-construction-related discovery must close no later than October 31, 2022. Dr. Osei's deposition has been noticed by Defendant's counsel, Brannick Howell & Stein LLP ("BHS"), with examining attorneys Victoria Ashworth (Lead Partner) and Daniel Kowalski (Associate Team Lead).
Deposition Notice. The deposition has been noticed by BHS pursuant to Federal Rule of Civil Procedure 30 and the Patent Local Rules of the Northern District of California. The notice designates the deposition as limited to claim-construction-related topics consistent with the Court's scheduling order.
Deposition Location. The deposition will be conducted at the offices of Calderwood & Harkness LLP, One Embarcadero Center, 34th Floor, San Francisco, California 94111.
THIS DOCUMENT IS PROTECTED BY THE ATTORNEY-CLIENT PRIVILEGE AND THE WORK PRODUCT DOCTRINE. IT SHOULD NOT BE DISCLOSED TO ANY PERSON OUTSIDE OF CALDERWOOD & HARKNESS LLP AND AUTHORIZED IN-HOUSE COUNSEL AT OPTIWAVE SEMICONDUCTOR CORP. THIS DOCUMENT MUST NOT BE PROVIDED TO DR. OSEI OR ANY OTHER WITNESS; IT IS FOR USE BY COUNSEL ONLY.
Right-click to update Table of Contents
I. WITNESS BACKGROUND AND QUALIFICATIONS
1.1 Professional Background
Dr. Amara Osei currently serves as Principal Engineer at Optiwave Semiconductor Corp., located at 2775 Augustine Drive, Suite 400, San Jose, California 95054. In this capacity, Dr. Osei leads the design and characterization of germanium-on-silicon photodetectors and plays a central role in the development of Optiwave's integrated silicon photonics platform. She is one of OWS's most senior technical contributors and reports directly to Dr. Kevin Fong, Chief Technology Officer and Co-Founder.
During the deposition, we anticipate that opposing counsel will begin with foundational background questions. Dr. Osei should be prepared to confirm the following:
• Current Title and Role: Principal Engineer, Optiwave Semiconductor Corp. Confirm the scope of her current responsibilities, including design, simulation, tape-out supervision, and post-fabrication characterization of germanium-on-silicon photodetector components integrated into OWS's silicon photonic integrated circuits ("PICs").
• Tenure at OWS: Confirm dates of employment and progression of roles from the time she joined OWS through her current position. Our understanding is that Dr. Osei joined OWS in approximately 2014–2015, initially as a Senior Design Engineer, and was promoted to Principal Engineer in or around 2019. Exact dates should be confirmed during the preparation session.
• Educational Background: Confirm the degrees she holds, the institutions from which they were conferred, and her fields of study. Our records indicate Dr. Osei holds a Ph.D. in Electrical Engineering (with a concentration in photonics and semiconductor device physics) and a B.S. in Electrical Engineering and Computer Science. The specific institutions should be confirmed with Dr. Osei. If she holds a master's degree, that should also be confirmed.
• Areas of Technical Specialization: Dr. Osei's areas of recognized expertise include: (i) germanium-on-silicon photodetector design, including p-i-n and avalanche photodiode architectures; (ii) integrated silicon photonics, with particular focus on receiver-side photonic components; and (iii) wavelength-division multiplexing on silicon platforms, including arrayed waveguide grating ("AWG") and echelle grating designs for coarse and dense WDM applications. These specializations are directly relevant to the subject matter of both asserted patents.
The purpose of establishing Dr. Osei's qualifications at the outset is twofold. First, to the extent that her testimony is later cited in claim construction briefing, it is important that the record reflect her deep expertise in the relevant fields. Second, her qualifications bear directly on her competence to testify about the understanding of a person of ordinary skill in the art ("POSITA"), which is a central issue in claim construction.
1.2 Inventorship
Dr. Osei is a named inventor on both patents at issue in this deposition:
U.S. Patent No. 10,891,234 ("the '234 Patent")
• Title: "Germanium-on-Silicon Photodetector with Graded Buffer Layer for Enhanced Responsivity"
• Filing Date: September 8, 2017
• Issue Date: January 14, 2021
• Named Inventors: Dr. Amara Osei and Dr. Kevin Fong
• Assignee: Optiwave Semiconductor Corp.
• Dr. Osei should be prepared to confirm her specific inventive contributions to the '234 Patent. Based on our understanding, Dr. Osei was the primary technical contributor responsible for the graded buffer layer architecture, including the design, simulation, and experimental validation of the SiGe buffer layer that is the central feature of the patented invention. Dr. Fong contributed to the overall device architecture and system-level integration concepts.
U.S. Patent No. 11,045,678 ("the '678 Patent")
• Title: "Multi-Channel Optical Transceiver with Integrated Wavelength-Division Multiplexing on Silicon"
• Filing Date: June 22, 2018
• Issue Date: June 29, 2021
• Named Inventors: Dr. Amara Osei, Dr. Priya Ramanathan, and Liwei Zhang
• Assignee: Optiwave Semiconductor Corp.
• Dr. Osei should be prepared to confirm her specific inventive contributions to the '678 Patent. Based on our understanding, Dr. Osei's contributions focused on the design and integration of the WDM demultiplexing element (specifically, the arrayed waveguide grating) on the silicon die, and her expertise in coupling the WDM element with the germanium photodetector array for multi-channel receiver functionality. Dr. Ramanathan contributed to the overall transceiver architecture and the transmitter-side laser integration concepts. Liwei Zhang contributed to the physical layout and mask design of the silicon photonic integrated circuit.
Dr. Osei should confirm that she has reviewed the patent specifications and claims of both the '234 Patent and the '678 Patent prior to the deposition. It is critical that she be familiar with the specific language of the asserted claims, particularly the disputed terms identified in the Joint Claim Construction and Prehearing Statement filed on October 14, 2022.
1.3 Prior Publications and Industry Experience
Dr. Osei should be prepared to discuss her publication record and prior industry experience. Opposing counsel may use her prior publications to explore whether she has used disputed claim terms in other contexts, which could be relevant to claim construction. Specifically:
• Publications in Integrated Photonics: Dr. Osei has authored or co-authored peer-reviewed publications in the field of integrated photonics, including journal articles and conference proceedings. During the preparation session, we should compile a list of her publications to identify any that use the terms "continuously varying," "graded buffer," "monolithically formed," or "monolithic integration." If any publication uses these terms in a manner inconsistent with OWS's proposed claim constructions, we need to address this proactively.
• Prior Employment: If Dr. Osei held positions at other companies, academic institutions, or government laboratories prior to joining OWS, she should be prepared to briefly describe those roles and her responsibilities. Any prior work on germanium photodetectors or WDM integration at a prior employer is particularly relevant.
• Experience with Pinnacle Semiconductor Foundry Co., Ltd. Processes: OWS's designs are fabricated by Pinnacle Semiconductor Foundry Co., Ltd. ("Pinnacle") on a 45nm silicon-on-insulator ("SOI") process. Dr. Osei should be prepared to discuss her familiarity with the Pinnacle process, including the epitaxial growth capabilities available through Pinnacle's platform, to the extent such information is not subject to confidentiality restrictions or trade-secret protections. This is relevant because the practical capabilities of the fabrication process bear on the meaning of terms like "continuously varying" in the context of real-world epitaxial growth.
1.4 Relationship to Other Named Inventors
Dr. Osei should be prepared to describe her working relationships with the other named inventors on the '234 and '678 Patents:
• Dr. Kevin Fong — Chief Technology Officer and Co-Founder of OWS. Dr. Fong is a co-inventor on the '234 Patent. Dr. Osei reports to Dr. Fong and has worked closely with him on photodetector design since joining OWS. Dr. Fong's own deposition (covering the '567, '234, and '890 Patents) is scheduled for later in October 2022. Coordination on overlapping topics—particularly the '234 Patent graded buffer layer invention—is essential to ensure consistency.
• Dr. Priya Ramanathan — Chief Executive Officer and Co-Founder of OWS. Dr. Ramanathan is a co-inventor on the '678 Patent. Dr. Osei should describe their collaborative work on the WDM transceiver architecture, while noting that Dr. Ramanathan's contributions were primarily on the transmitter side and overall system architecture.
• Liwei Zhang — Design Engineer at OWS. Mr. Zhang is a co-inventor on the '678 Patent. Dr. Osei should describe his contributions to the physical layout and mask design of the silicon photonic integrated circuit. Mr. Zhang is a more junior member of the team.
II. PREPARATION NOTES AND KEY THEMES
2.1 Purpose of Deposition in Claim Construction Context
This deposition is being conducted during the claim construction discovery phase of the litigation. It is critical that Dr. Osei—and defending counsel—understand the limited scope of this deposition:
• The deposition is limited to claim-construction-related topics pursuant to the Court's scheduling order and the Patent Local Rules of the Northern District of California. Specifically, the focus will be on Dr. Osei's understanding—as a named inventor and person of skill in the art—of the meaning of disputed claim terms, the state of the art at the time of the inventions, and the prosecution history of the '234 and '678 Patents.
• This is not a merits deposition on infringement or invalidity. Questions directed to whether specific Lumos products infringe the asserted claims, or whether the asserted claims are invalid over prior art, are beyond the scope of this deposition. Defending counsel should be prepared to object to and, if necessary, instruct the witness not to answer questions that venture into infringement or validity territory beyond what is relevant to claim construction.
• However, the line between claim construction testimony and infringement/validity testimony can be blurry. Opposing counsel may attempt to elicit testimony about the scope of the claims in a way that effectively constitutes infringement or validity testimony. Defending counsel should be vigilant about this.
2.2 Key Objectives for Our Side (Defensive Preparation)
Although this deposition is being taken by Defendant's counsel (BHS) and Dr. Osei will be a hostile witness from BHS's perspective, the fact that she is OWS's own inventor means that her testimony will be closely scrutinized and potentially used against OWS if she makes damaging concessions. Our key objectives for this deposition are:
1. Support OWS's broader proposed claim constructions for the disputed terms in the '234 and '678 Patents. Dr. Osei's testimony should be consistent with and supportive of OWS's proposed constructions as set forth in OWS's Preliminary and Responsive Claim Construction Contentions.
2. Establish that a POSITA would understand the disputed claim terms consistent with OWS's proposed constructions. As a named inventor and highly qualified engineer in the relevant field, Dr. Osei is well-positioned to testify about how a POSITA would understand the claim terms in the context of the patent specifications and the state of the art.
3. Avoid concessions that would support Lumos's narrow constructions. This is the most critical defensive objective. BHS will attempt to elicit admissions from Dr. Osei that support Lumos's proposed narrow constructions—particularly that "continuously varying" excludes step-graded designs and that "monolithically formed" excludes all forms of heterogeneous integration.
4. Avoid statements that could create or exacerbate prosecution history estoppel issues. This concern is particularly acute for the '678 Patent, where "monolithically formed" was added by amendment during prosecution and the applicant's remarks expressly distinguished heterogeneous integration approaches. Any testimony from Dr. Osei that confirms or amplifies the scope of the prosecution disclaimer could severely damage OWS's claim construction position.
5. Preserve OWS's ability to use Dr. Osei as a trial witness. Nothing Dr. Osei says at this deposition should undermine her credibility or usefulness as a witness at the Markman hearing or at trial.
2.3 Anticipated Lines of Defense Questioning (Brannick Howell & Stein)
Based on BHS's claim construction positions and the composition of their litigation team, we anticipate the following lines of questioning:
• Examining Counsel. We expect Daniel Kowalski, Associate Team Lead at BHS, to lead the substantive questioning. Kowalski holds a Ph.D. in Optics from the University of Rochester and is technically sophisticated. He will likely attempt to engage Dr. Osei in technical discussions designed to elicit admissions on narrow meanings of the disputed claim terms. He is known to be methodical and persistent, often returning to the same topic from multiple angles. Victoria Ashworth, Lead Partner at BHS, may participate in questioning or may attend in a supervisory capacity. Ashworth is a seasoned patent trial lawyer who is less technically oriented but highly effective at framing questions to create a favorable record.
• Technical Consultant. We understand that Dr. Raymond Ng, Of Counsel and Technical Consultant to BHS (formerly a professor of photonics at the University of Colorado Boulder), may attend the deposition in a consulting capacity. Dr. Ng may suggest technical lines of questioning to Kowalski in real time.
• Focus Areas. Based on the disputed claim terms and BHS's stated positions, we anticipate the following specific lines of questioning:
(a) '234 Patent — "Continuously Varying" vs. "Step-Graded." Expect sustained questioning on the distinction between "continuously varying" and "step-graded" buffer layers. Kowalski will likely present Dr. Osei with textbook definitions, academic literature, and the patent specification to argue that "continuously varying" has a clear, narrow meaning that excludes discrete step changes in germanium concentration. He may present concentration profile diagrams showing smooth curves versus staircase functions and ask Dr. Osei to agree that they are fundamentally different.
(b) '678 Patent — "Monolithically Formed" and Heterogeneous Integration. Expect detailed questioning on the meaning of "monolithically formed" and whether it excludes hybrid integration, flip-chip bonding, micro-transfer printing, and other heterogeneous integration techniques. Kowalski will almost certainly read from the prosecution history remarks distinguishing heterogeneous integration and ask Dr. Osei to confirm the scope of those remarks.
(c) Prosecution History Amendments. Expect questions about the reasons claims were narrowed during prosecution, particularly for the '678 Patent. BHS will attempt to establish that the amendments and accompanying remarks represent binding admissions about the scope of the claims.
(d) Specification vs. Claim Scope. Expect questions designed to establish that the claims should be limited to the specific embodiments described in the specification, particularly where the specification describes alternative approaches (e.g., step-graded buffers) that are not reflected in the claim language.
III. THE '234 PATENT — DETAILED TOPIC AREAS
3.1 Overview of the Invention
The '234 Patent is titled "Germanium-on-Silicon Photodetector with Graded Buffer Layer for Enhanced Responsivity." The core inventive concept is a germanium-on-silicon photodetector that incorporates a graded buffer layer between the silicon substrate and the germanium absorber layer. The graded buffer layer serves to reduce threading dislocation density at the germanium-silicon interface, which in turn enhances the optical responsivity and reduces dark current in the photodetector. This is a well-recognized problem in the germanium-on-silicon integration field: the approximately 4.2% lattice mismatch between germanium and silicon leads to a high density of threading dislocations when germanium is grown directly on silicon, degrading device performance.
Key patent details:
• Filing Date: September 8, 2017
• Issue Date: January 14, 2021
• Named Inventors: Dr. Amara Osei and Dr. Kevin Fong
• Assignee: Optiwave Semiconductor Corp.
• Asserted Claims: Claims 1, 3, 5, 8, and 11
Dr. Osei should confirm her role in developing the specific buffer layer architecture that is the subject of the '234 Patent. Based on our understanding, she was the primary technical contributor who designed, simulated, and experimentally validated the graded SiGe buffer layer structure. She should be able to explain the technical problem that the invention addresses (lattice mismatch, threading dislocations, reduced responsivity), the solution the invention provides (a graded buffer layer that gradually transitions the lattice constant from silicon to germanium), and the benefits achieved (reduced dislocation density, enhanced responsivity, lower dark current).
3.2 Key Disputed Term: "Graded Buffer Layer Having a Continuously Varying Germanium Concentration"
This is the most significant disputed term for the '234 Patent and will be the primary focus of BHS's questioning of Dr. Osei regarding this patent. The competing constructions are:
The critical distinction between these constructions is whether "continuously varying" encompasses step-graded buffer designs or is limited to continuously graded (i.e., smooth, uninterrupted gradient) designs. OWS's broader construction would cover both continuously graded and step-graded buffer architectures, while Lumos's narrow construction would exclude step-graded designs entirely. This distinction is outcome-determinative for the infringement analysis because Lumos's accused photodetector uses a step-graded buffer design.
Preparation Points for Dr. Osei:
(a) What a POSITA would understand. Dr. Osei should be prepared to testify that a person of ordinary skill in the art would understand "continuously varying germanium concentration" to describe a buffer layer in which the germanium concentration changes across the layer thickness, as opposed to an abrupt, single-step junction between silicon and germanium. The term "continuously varying" is used in the specification to contrast with abrupt interfaces, not to draw a sharp line between smooth gradients and fine step-graded approaches.
(b) Practical realities of epitaxial growth. Dr. Osei should explain that in practice, even nominally "continuously" graded buffers are achieved through incremental changes in process parameters during epitaxial growth (whether by molecular beam epitaxy ("MBE") or chemical vapor deposition ("CVD")). The gas-phase composition in a CVD reactor, for example, is adjusted in finite steps by the mass flow controllers. At the atomic scale, there is always some degree of discretization. Thus, the distinction between a "continuously" graded buffer and a "finely step-graded" buffer is one of degree and growth parameter resolution, not a fundamental physical difference.
(c) Contrast with abrupt junctions. Dr. Osei should emphasize that the inventive concept of the '234 Patent is the use of a graded buffer—as opposed to an abrupt junction—to reduce threading dislocation density. The term "continuously varying" in the specification is used to describe this inventive concept in contrast to abrupt junctions. It is not used as a term of exclusion directed at step-graded approaches, which achieve the same fundamental objective of gradually transitioning the lattice constant.
(d) Specification describes both approaches. The '234 Patent specification describes embodiments with both continuously graded and step-graded SiGe buffer layers. This is significant because it demonstrates that the inventors contemplated both approaches as within the scope of the inventive concept. However, Dr. Osei should be aware that only "continuously varying" language appears in the claims themselves—this is the basis of Lumos's argument that the claims exclude step-graded designs. See Section 3.3 below for detailed preparation on this point.
3.3 Specification Embodiments — Step-Graded vs. Continuously Graded
This is a high-risk area. The '234 Patent specification describes embodiments with both continuously graded and step-graded SiGe buffer layers. The claims, however, use only "continuously varying germanium concentration" language. Lumos will seize on this discrepancy to argue that the claims were deliberately drafted to cover only the continuously graded embodiment, and that the step-graded embodiment was left unclaimed.
Preparation Points for Dr. Osei:
(a) Why the specification describes both embodiments. Dr. Osei should be prepared to explain that the specification describes both continuously graded and step-graded embodiments to illustrate the breadth of the inventive concept. The specification is intended to provide a full disclosure of the invention and its various implementations. The fact that both embodiments are described demonstrates that the inventors understood their invention to encompass the general concept of a graded buffer layer, whether implemented as a continuous gradient or as a series of discrete steps.
(b) Understanding of claim scope. If asked whether the claims were intended to cover only the continuously graded embodiment or both, Dr. Osei should be cautious. She should testify that, as an inventor, she understood the invention to encompass the general concept of grading the germanium concentration across the buffer layer, and that her understanding is that "continuously varying" in the context of epitaxial growth encompasses the range of graded approaches, including finely step-graded buffers. However, she should not purport to offer legal opinions on claim scope.
(c) Caution — Defer to patent counsel on claim drafting. Dr. Osei should not speculate on prosecution strategy or the reasons why specific claim language was chosen. If asked why the claims use "continuously varying" rather than broader language that explicitly encompasses step-graded designs, she should defer to the patent attorney who drafted the claims—Ted Marchetti, VP of intellectual Property at OWS (USPTO Reg. No. 58,304)—and outside prosecution counsel. She should state that she is an inventor, not a patent attorney, and that claim drafting decisions were made by the IP professionals.
(d) Key risk. Lumos will try to elicit testimony that "continuously varying" was deliberately chosen in the claims to distinguish from step-graded designs. This would be devastating to OWS's broader construction. Dr. Osei must be prepared for this line of questioning and must not concede this point. If pressed, she should reiterate that she was not involved in the claim drafting process and cannot speak to the reasons for specific word choices in the claims.
3.4 State of the Art in Germanium Photodetectors (as of September 2017 Filing Date)
Dr. Osei will likely be asked about the state of the art in germanium-on-silicon photodetectors as of the filing date of the '234 Patent (September 8, 2017). This testimony is relevant to claim construction because the understanding of a POSITA is assessed in light of the state of the art.
Known Approaches at the Time:
(a) Direct epitaxial growth of Ge on Si. This approach involves growing a germanium layer directly on the silicon substrate without an intermediate buffer layer. Due to the 4.2% lattice mismatch, this approach results in a high density of threading dislocations (typically on the order of 10⁸–10⁹ cm⁻²), which degrades device performance through increased dark current and reduced responsivity.
(b) Step-graded SiGe buffers. Many companies and research groups used step-graded buffer layers with discrete composition steps. A common approach involved four-step SiGe buffers with germanium fractions of approximately 0.25, 0.50, 0.75, and 1.0. This approach reduces threading dislocation density to approximately 10⁶–10⁷ cm⁻² by distributing the lattice mismatch across multiple interfaces. Notably, a four-step SiGe buffer with germanium fractions of 0.25, 0.50, 0.75, and 1.0 is exactly the design used in Lumos's accused photodetector. Dr. Osei should be aware of this but should not volunteer this information.
(c) Continuously graded SiGe buffers. Some researchers employed continuously graded buffers in which the germanium concentration is ramped smoothly (or quasi-smoothly) from silicon to germanium. This approach can achieve even lower threading dislocation densities (10⁵–10⁶ cm⁻²) and smoother surface morphology.
(d) Selective area growth. Another approach involves selective epitaxial growth of germanium in predefined trenches or mesas, using the trench geometry to trap threading dislocations at the sidewalls (aspect ratio trapping technique).
Preparation Points:
• Dr. Osei should be prepared to discuss the advantages of continuously graded buffers over step-graded buffers (smoother lattice transition, lower threading dislocation density, higher responsivity) without conceding that "continuously varying" was a recognized term of art in the field that specifically and exclusively referred to smooth gradients as distinguished from step-graded approaches.
• If asked whether "continuously varying" and "step-graded" were understood in the art to be different things, Dr. Osei should acknowledge that these terms describe different idealized profiles, but explain that in practice, the distinction is one of degree. A "continuously" graded buffer grown by CVD involves incremental changes in precursor gas flow rates; a "step-graded" buffer with many fine steps approaches a continuous gradient. The two represent points on a continuum rather than fundamentally different categories.
• Dr. Osei should be prepared to identify specific publications, conference proceedings, or industry reports from the relevant time period that she was aware of or relied upon in developing the '234 Patent invention. During the preparation session, we should review her publication history and any references cited in the '234 Patent specification to ensure consistency.
3.5 Prosecution History of the '234 Patent
The prosecution history of the '234 Patent is relatively clean from OWS's perspective:
• The '234 Patent was allowed on the first office action after an examiner interview. No substantive amendments narrowing the claims were made during prosecution. This means there is no prosecution history estoppel issue for the '234 Patent—unlike the '678 Patent, where the "monolithically formed" limitation was added by amendment.
• The examiner interview was conducted by Ted Marchetti and outside prosecution counsel. Dr. Osei was not involved in the prosecution of the patent and did not attend the examiner interview.
• Although there is no prosecution history estoppel issue, Lumos may argue that the claim language itself is clear and unambiguous and that "continuously varying" has a plain and ordinary meaning—namely, a smooth, uninterrupted gradient without step changes. This is a claim construction argument based on intrinsic evidence (the claim language and specification), not prosecution history estoppel.
Preparation Points for Dr. Osei:
• If asked about the examiner interview, Dr. Osei should confirm that she was not involved in prosecution decisions. Those were handled by Ted Marchetti (VP of IP, Reg. Pat. Atty., USPTO Reg. No. 58,304) and outside prosecution counsel. She should not speculate about what was discussed during the examiner interview.
• If asked whether she reviewed or approved the claims before they were filed, she should explain that she provided technical input to the patent attorney but that the claim language was drafted by the patent professionals. She may have reviewed a draft of the specification or claims, but she did not make independent decisions about claim language.
3.6 Relationship to Prior Art
The following prior art reference is relevant to Dr. Osei's deposition testimony regarding the '234 Patent:
• European Patent Application EP 3,234,567 A1 (published August 2, 2017; applicant: Nordlicht Photonische GmbH). This reference was cited against the '234 Patent during prosecution under 35 U.S.C. § 102(a)(2). The Nordlicht reference discloses a germanium photodetector with a buffer layer, but—as argued by OWS during prosecution—it does not teach or suggest the specific graded buffer architecture claimed in the '234 Patent.
Preparation Points for Dr. Osei:
• Dr. Osei should be prepared to address her awareness of the Nordlicht reference at the time of invention. Did she know of Nordlicht's work? Did she review the Nordlicht reference during the development of the '234 Patent invention? If so, did it influence her design decisions?
• If Dr. Osei was not aware of the Nordlicht reference prior to the patent prosecution, she should simply say so. If she was aware of it, she should explain how the '234 Patent invention differs from the Nordlicht approach.
• Note that the Nordlicht reference was published on August 2, 2017—approximately one month before the '234 Patent filing date of September 8, 2017. It is possible that Dr. Osei was not aware of it at the time of filing. This should be confirmed during the preparation session.
3.7 The Accused Lumos Photodetector Design
While this deposition is scoped to claim construction and not infringement, it is important for Dr. Osei to be aware of the accused Lumos design so that she understands the stakes of her testimony on the disputed claim terms:
• Lumos's Photodetector Design. Lumos's germanium-on-silicon photodetector uses a step-graded SiGe buffer layer with four discrete composition steps at germanium fractions of approximately 0.25, 0.50, 0.75, and 1.0. This is a classic four-step staircase profile.
• Relevance to Claim Construction. If OWS's broader construction of "continuously varying germanium concentration" is adopted, Lumos's step-graded design may fall within the scope of the asserted claims. If Lumos's narrow construction is adopted (requiring a monotonically increasing, smooth gradient without step changes), Lumos's step-graded design would be excluded from the claim scope.
• Preparation Points. Dr. Osei should not volunteer information about Lumos's specific products or offer opinions on whether they infringe. If asked about Lumos's products in the context of claim construction, she should redirect her testimony to the meaning of the claim terms to a POSITA, not to whether specific accused products fall within the claim scope. If opposing counsel persists in asking about infringement, defending counsel should object and, if necessary, instruct the witness not to answer.
IV. THE '678 PATENT — DETAILED TOPIC AREAS
4.1 Overview of the Invention
The '678 Patent is titled "Multi-Channel Optical Transceiver with Integrated Wavelength-Division Multiplexing on Silicon." The core inventive concept is a multi-channel optical transceiver that integrates a WDM element (such as an arrayed waveguide grating or echelle grating) monolithically on a single silicon die, enabling higher-density optical interconnects for data center and telecommunications applications. At the time of the invention, the industry trend was toward increasing data rates and channel counts in optical interconnects, and the '678 Patent addressed the need for compact, low-cost, high-performance WDM integration on a silicon photonics platform.
Key patent details:
• Filing Date: June 22, 2018
• Issue Date: June 29, 2021
• Named Inventors: Dr. Amara Osei, Dr. Priya Ramanathan, and Liwei Zhang
• Assignee: Optiwave Semiconductor Corp.
• Asserted Claims: Claims 1, 2, 6, 9, and 14
Dr. Osei should confirm her specific contributions to the WDM integration architecture. Based on our understanding, her primary contributions were: (i) the design of the integrated WDM demultiplexing element (specifically, an arrayed waveguide grating) in the silicon nitride waveguide layer on the silicon die; (ii) the optical coupling scheme between the WDM element and the germanium photodetector array; and (iii) the optimization of the WDM passband characteristics for compatibility with standard CWDM/DWDM wavelength grids. Dr. Ramanathan contributed the overall transceiver system architecture and the transmitter-side concepts. Liwei Zhang contributed the physical layout and mask design.
4.2 Key Disputed Term: "Integrated Wavelength-Division Multiplexing Element Monolithically Formed on a Single Silicon Die"
This is the most significant—and most dangerous—disputed term for the '678 Patent. The competing constructions are:
The critical distinction is whether "monolithically formed" is a process-focused term (fabricated as part of the same fabrication process) or a structural/exclusionary term (formed within the same silicon substrate, excluding all forms of heterogeneous integration). OWS's broader construction focuses on the fabrication process and would not exclude optical engines that also incorporate heterogeneously integrated components (such as III-V gain elements) alongside the monolithically formed WDM element. Lumos's narrow construction explicitly excludes hybrid integration, flip-chip bonding, and any heterogeneous integration technique, which would effectively exclude Lumos's LumosBridge 800G platform from the claim scope because that platform uses micro-transfer-printed III-V gain elements bonded to the silicon PIC—even though the WDM element itself is monolithically fabricated in the silicon waveguide layer.
Preparation Points for Dr. Osei:
(a) What a POSITA would understand "monolithically formed" to mean. Dr. Osei should be prepared to testify that a person of ordinary skill in the silicon photonics field would understand "monolithically formed" to mean that the component in question (here, the WDM element) is fabricated on or within the silicon die as part of the semiconductor fabrication process, rather than being a separate, discrete component that is assembled onto the die after fabrication. The term refers to the WDM element's relationship to the silicon die, not to the entirety of the optical engine.
(b) Usage in the photonics literature. Dr. Osei should explain that "monolithic" in the photonics literature is commonly used to describe integration on a single substrate, even when the fabrication process involves multiple materials (e.g., silicon and silicon nitride) or multiple process steps. The term does not, in ordinary usage in the field, require that every component on the die be fabricated in a single, undifferentiated process step.
(c) Key distinction intended by the inventors. Dr. Osei should testify that the key distinction the inventors intended was between a fully integrated, on-chip WDM element (patterned in the silicon or silicon nitride waveguide layer during fabrication) versus a separate, off-chip WDM component (such as a discrete thin-film filter or fiber Bragg grating) that would need to be optically coupled to the silicon PIC through external packaging. The invention solved the problem of integrating the WDM function onto the silicon die, eliminating the need for a separate optical component and the associated coupling losses, packaging complexity, and cost.
(d) The claim addresses the WDM element, not the entire optical engine. A critical point: the disputed claim term refers to the "integrated wavelength-division multiplexing element" being "monolithically formed on a single silicon die." It does not require that every component of the optical engine or transceiver be monolithically formed. Other components—such as laser sources or optical amplifiers—may be integrated using different techniques (including heterogeneous integration of III-V materials) without affecting whether the WDM element itself is monolithically formed.
4.3 Prosecution History of the '678 Patent — Critical Estoppel Issues
The prosecution history of the '678 Patent is the most significant risk area in this deposition. Unlike the '234 Patent, which was allowed on the first office action, the '678 Patent underwent a contested prosecution with three office actions, and the claim language was amended in ways that Lumos will argue create prosecution history estoppel.
Key Prosecution Events:
(a) Three Office Actions. The '678 Patent application received three office actions from the USPTO before allowance. During prosecution, the examiner cited prior art references disclosing optical transceivers with WDM elements integrated using hybrid or heterogeneous techniques.
(b) Addition of "Monolithically Formed." The term "monolithically formed" was added to the claims by amendment to overcome the examiner's rejection. The original claims used broader language (the specific original language should be confirmed by reviewing the prosecution file history, but it is our understanding that the original claims referred to an "integrated WDM element" without the "monolithically formed" qualifier).
(c) Applicant's Prosecution Remarks. In the remarks accompanying the amendment, applicant's counsel stated: "The present invention is distinguished from heterogeneous integration approaches." This statement was made in the context of distinguishing the claimed invention from the specific prior art cited by the examiner, which disclosed a hybrid integration approach.
Prosecution History Estoppel Risk:
This prosecution history creates a significant estoppel risk. Under the doctrine of prosecution history estoppel, a patentee who narrows claim scope during prosecution to overcome a prior art rejection may be estopped from recapturing the surrendered scope. Lumos will argue that by adding "monolithically formed" and expressly distinguishing from "heterogeneous integration approaches," OWS surrendered any claim scope covering heterogeneous integration techniques.
The strength of Lumos's estoppel argument depends on the breadth of the surrender. If the applicant's remarks are read broadly, they could be interpreted as disclaiming all forms of heterogeneous integration. If they are read narrowly, they may be limited to the specific type of heterogeneous integration disclosed in the cited prior art.
Preparation Points for Dr. Osei:
(a) Expect questions about the prosecution remarks. Lumos's counsel (likely Kowalski) will almost certainly read the applicant's prosecution remarks verbatim and ask Dr. Osei to confirm that the invention excludes heterogeneous integration. This is a predictable and dangerous line of questioning.
(b) Clarify her role. Dr. Osei should state clearly that she did not draft the prosecution remarks and is not a patent attorney. The prosecution of the '678 Patent was handled by Ted Marchetti (VP of IP, USPTO Reg. No. 58,304) and outside prosecution counsel. She did not participate in prosecution strategy decisions and cannot speak to the legal significance or intended scope of the remarks.
(c) Her understanding as an engineer. Dr. Osei should state that her understanding of "monolithically formed" as an engineer may differ from the legal arguments made during prosecution. As an engineer, she understands "monolithically formed" to describe the fabrication of the WDM element on the silicon die, which is a technical description of the manufacturing process. She does not interpret the term as a legal conclusion about what other integration techniques may or may not be present elsewhere on the same die.
(d) Do NOT concede that "monolithically formed" necessarily excludes all forms of hybrid or heterogeneous integration. This is the single most important defensive point for the '678 Patent. If Dr. Osei concedes this point, it will severely undermine OWS's claim construction position. She should avoid categorical statements about what "monolithically formed" excludes.
(e) Distinguish between what was necessary to overcome the specific prior art and the full scope of the claim language. The prosecution remarks were made in the context of distinguishing a specific prior art reference that disclosed a particular type of heterogeneous integration. The fact that the applicant distinguished from that specific approach does not necessarily mean that all forms of heterogeneous integration are excluded from the claim scope. Dr. Osei should not try to make this legal argument herself—it is a legal point for briefing—but she should not foreclose it by agreeing that the prosecution remarks have the broad meaning Lumos attributes to them.
(f) Defer to Ted Marchetti. If pressed on prosecution strategy, Dr. Osei should state that questions about prosecution strategy and the intent behind prosecution remarks should be directed to Ted Marchetti, who handled the prosecution. She was not involved in those decisions and cannot speak to them.
4.4 Specification Embodiments for the '678 Patent
The '678 Patent specification provides additional context for the meaning of "monolithically formed":
(a) Preferred Embodiment. The specification describes the preferred embodiment as a WDM element (specifically, an arrayed waveguide grating or echelle grating) fabricated within the same CMOS-compatible process flow on the silicon die. The WDM element is patterned in the silicon or silicon nitride waveguide layer using standard lithographic techniques, and it is integrated with the germanium photodetector array and the optical coupling structures in a single, monolithic fabrication sequence.
(b) Background Section — Prior Art Approaches. The specification's background section discusses prior art approaches to WDM integration, including hybrid integration, flip-chip bonding, and heterogeneous bonding. These approaches are described as disadvantageous due to increased packaging complexity, higher cost, reduced reliability, and optical coupling losses. Lumos will argue that the specification's criticism of these prior art approaches supports a narrow construction that excludes them from the claim scope.
(c) OWS's Position. OWS's position is that the specification's criticism of prior art hybrid approaches describes the state of the art and motivates the invention—it does not mean that the claims are limited to only the preferred embodiment. It is well-established that claims are not limited to the preferred embodiment. The specification's description of disadvantages of hybrid approaches explains why the monolithic integration approach of the invention is advantageous, but it does not constitute a disclaimer of claim scope.
Preparation Points for Dr. Osei:
Dr. Osei should be prepared to explain that the specification's discussion of prior art hybrid approaches is background information that describes the state of the art and motivates the invention. She should not concede that the claims are limited to the specific embodiment described in the specification. If asked whether the claims exclude hybrid integration approaches, she should state that the claims describe the WDM element as being monolithically formed on the silicon die, and that this is a description of how the WDM element is fabricated—it does not address what other components on the die may or may not involve other integration techniques.
4.5 State of the Art in WDM Integration on Silicon (as of June 2018 Filing Date)
Dr. Osei should be prepared to discuss the state of the art in WDM integration on silicon as of the filing date of the '678 Patent (June 22, 2018):
(a) Monolithic Integration of WDM on Silicon. By 2018, it was well-established that WDM multiplexing and demultiplexing elements—such as arrayed waveguide gratings and echelle gratings—could be fabricated monolithically on silicon using silicon or silicon nitride waveguides. This was a mature technology with demonstrated performance in both CWDM and DWDM applications. Multiple silicon photonics foundries offered process design kits (PDKs) that included AWG and echelle grating building blocks.
(b) Hybrid Integration with III-V Materials. At the same time, the industry was increasingly adopting hybrid and heterogeneous integration techniques for components that could not be efficiently realized in silicon—particularly laser sources and optical amplifiers, which require III-V semiconductor materials (such as InP or InGaAsP) for efficient light emission. Common techniques included wafer bonding, flip-chip bonding, and micro-transfer printing of III-V gain elements onto the silicon PIC.
(c) Distinction Between WDM Elements and Gain Elements. A critical point that Dr. Osei should make clear in her testimony is the distinction between the WDM element (which the '678 Patent claim specifically addresses) and gain elements (lasers, amplifiers) that may require III-V materials. In the state of the art as of 2018, WDM multiplexing/demultiplexing elements were increasingly fabricated monolithically on silicon, while gain elements were typically heterogeneously integrated. The '678 Patent claims address the WDM element, not the gain elements.
Preparation Points for Dr. Osei:
Dr. Osei should be prepared to testify about the state of the art in a manner that supports the distinction between the WDM element (monolithically formed) and other components of the optical engine (which may be integrated using other techniques). She should not conflate the WDM element with the laser or amplifier components, and she should not testify in a way that suggests the entire optical engine must be monolithically formed for the claim to be satisfied. The claim language specifically refers to the "integrated wavelength-division multiplexing element" being "monolithically formed"—not the entire transceiver.
4.6 Relationship to Accused Lumos Products
• LumosBridge 800G Optical Engine. Lumos's LumosBridge 800G optical engine, which is the primary accused product for the '678 Patent, uses micro-transfer-printed III-V gain elements (laser sources) bonded to the silicon photonic integrated circuit. However, our understanding is that the WDM demultiplexing element in the LumosBridge 800G is fabricated monolithically in the silicon nitride waveguide layer on the silicon die—it is not a separately fabricated component that is bonded or attached to the die.
• The Central Question. The question for claim construction is whether "monolithically formed on a single silicon die" can be satisfied when the WDM element itself is monolithically fabricated, even if other components on the same die (such as the III-V gain elements) are heterogeneously integrated. OWS says yes; Lumos says no.
• Preparation Points for Dr. Osei. Dr. Osei should focus her testimony on the claim term "monolithically formed" as it applies to the WDM element specifically, not to the entire optical engine. If asked about Lumos's products, she should redirect to the meaning of the claim terms to a POSITA. She should not offer infringement opinions.
V. CROSS-EXAMINATION RISK AREAS AND PROTECTIVE INSTRUCTIONS
This section identifies the most significant risks that may arise during cross-examination by BHS counsel and sets forth protective instructions for Dr. Osei. These instructions should be reviewed with Dr. Osei during the preparation session and reinforced immediately before the deposition.
5.1 "Continuously Varying" — Risk of Conceding Narrow Meaning
Risk: Dr. Osei is a highly trained scientist and engineer. In her normal professional usage, she may naturally use "continuously varying" in its strict mathematical or physical sense—meaning a smooth, uninterrupted gradient that is a continuous function (in the mathematical sense) of position. This is the intuitive meaning of the term, and it is the meaning that Lumos advocates. If Dr. Osei testifies that "continuously varying" means a smooth gradient without step changes, she will have effectively conceded Lumos's proposed construction.
Protective Instructions:
(a) Remind Dr. Osei that in the context of epitaxial growth of semiconductor buffer layers, even nominally "continuous" gradients are achieved through incremental process steps. The precursor gas flow rates in a CVD reactor are adjusted by mass flow controllers that operate in finite increments. The result is a composition profile that, at sufficiently high magnification, shows small steps. The distinction between "continuously varying" and "finely step-graded" is therefore one of degree, not kind.
(b) If asked whether "continuously varying" means "without step changes," Dr. Osei should explain that a POSITA would understand the term in the context of real-world epitaxial growth processes, where perfectly smooth composition gradients do not exist at the atomic scale. The term describes a general grading concept, not a mathematically precise requirement for a continuous function.
(c) Dr. Osei should avoid using the terms "continuously varying" and "step-graded" as though they are mutually exclusive categories. Instead, she should describe them as points on a spectrum of graded buffer approaches.
(d) If Kowalski presents textbook definitions of "continuous function" or "continuous variation" and asks Dr. Osei to agree that the term has a specific mathematical meaning, she should acknowledge that the mathematical definition exists but explain that in the context of semiconductor fabrication, the term is understood more broadly. Patent claim terms are not mathematical abstractions; they describe physical structures and processes.
5.2 Prosecution Remarks and the '678 Patent
Risk: This is the highest-risk area of the deposition. Lumos's counsel will almost certainly read the applicant's prosecution remarks verbatim: "The present invention is distinguished from heterogeneous integration approaches." Kowalski will then ask Dr. Osei to confirm that the invention excludes heterogeneous integration. If she agrees, it will significantly strengthen Lumos's prosecution history estoppel argument.
Protective Instructions:
(a) Dr. Osei should state clearly and unambiguously that she did not draft the prosecution remarks and is not a patent attorney. The prosecution of the '678 Patent was handled by Ted Marchetti and outside prosecution counsel. She was not involved in the prosecution strategy and cannot speak to the legal significance of the remarks.
(b) She should state that her understanding of the invention, as an engineer and inventor, is that the '678 Patent addresses the integration of the WDM element on the silicon die. The "monolithically formed" limitation describes how the WDM element is fabricated. She understood the invention to solve a different problem than what the cited prior art presented.
(c) She should NOT try to reinterpret the prosecution remarks or argue that they mean something different from what they appear to say. If she attempts to reinterpret the remarks, she will appear evasive and will likely be pressed further, potentially making things worse. The better approach is simply to note that she was not involved in prosecution and cannot speak to the intent or scope of the remarks.
(d) She should NOT concede that "monolithically formed" necessarily excludes all forms of hybrid or heterogeneous integration. If pressed, she should repeat that she understands the term to describe the fabrication of the WDM element, and that questions about the legal scope of the claims should be directed to patent counsel.
(e) If Kowalski asks: "Do you agree with the statement made by your patent attorney during prosecution that the invention is distinguished from heterogeneous integration approaches?" Dr. Osei should respond that she was not involved in prosecution, she did not author those remarks, and she cannot agree or disagree with a legal argument she did not make.
5.3 Specification Descriptions vs. Claim Scope
Risk: Lumos will attempt to limit the scope of the asserted claims to the preferred embodiments described in the specifications. For the '234 Patent, this means arguing the claims cover only the continuously graded embodiment (not the step-graded embodiment also described in the specification). For the '678 Patent, this means arguing the claims exclude the hybrid integration approaches criticized in the background section.
Protective Instructions:
(a) Dr. Osei should not testify that the claims are limited to any single embodiment described in the specification. If asked, she should note that the specification describes exemplary embodiments to illustrate the invention, but that the claims define the scope of the invention.
(b) She should not volunteer interpretations of claim scope. She is a fact witness and technical expert, not a legal expert on claim construction. If pressed on the legal question of what the claims cover, she should defer to the attorneys and the Court.
(c) If Kowalski asks: "Isn't it true that the specification only describes one type of graded buffer in the detailed description—a continuously graded buffer?" Dr. Osei should correct the premise if inaccurate. Our understanding is that the '234 Patent specification describes both continuously graded and step-graded embodiments. She should point to the relevant passages in the specification.
5.4 Awareness of Lumos's Products